Eradicating data discrepancies to support people living with HIV/AIDS in Tanzania

After building skills working with and cleaning data, a PEPFAR implementing partner in the Southern Highlands transforms its data management and reporting, reduces inefficiencies, and improves resource allocation.
Mbeya City, Tanzania

Background

Joseph provides community-based HIV/AIDS services (HIV/AIDS) with KIHUMBE, a civil society organization operating in four districts in Mbeya Region Mbeya City, Mbeya Rural, Rungwe and Lusekelo.

As a PEPFAR implementing partner, KIHUMBE works closely with 240 community volunteers to provide HIV testing services, psychosocial support to people living with HIV (PLHIV), biomedical services,medical services, and social / vocational services (including linking the PLHIV with financial facilities for soft loans/grants). At least 20,000 people living with HIV/AIDS are accessing KIHUMBE services in the region.

Problem

“Data ilikuwa chafu sana” [“Data was so dirty”] – Joseph

At KIHUMBE, Joseph and his team of 240 volunteers across the region produce large amounts of data on daily basis. These CBHS volunteers are the core producers of data pertaining the community based HIV/AIDS services—but most of them are non-data specialists who often care less on ‘what’ they report to a higher level. As a result, routine datasets on program implementation are often incomplete, of questionable quality, and not easily accessible for use in analysis and interpretation. As Joseph explained:

“Data that was coming from the field had a lot of typos, [and it was] incomplete and unreliable. This compelled us to invest more time and resources to follow up with to volunteers for clarifications.

Poor data quality adversely affected KIHUMBE by:

• Causing inefficiencies in CBHS processes which depended on data, like creating reports. These inefficiencies resulted in very expensive rework efforts to “fix” the data in order to meet the requirements of various program processes.
• Leading to poor decisions for program implementation based on flawed data.
• Fomenting mistrust with the funder as time and money was wrongly utilized.

Solution

In July 2017, Data Zetu hosted a data literacy and data management training for organizations tackling HIV/AIDS in Mbeya Region, in collaboration with the dLab. After attending that training, Joseph in turn oriented the 240 other CBHS community volunteers across the region, as well as six KIHUMBE field officers between October and November 2017. These trainings instituted data quality control mechanisms at the hyper local level across KIHUMBE’s areas of work.

Process

Data Zetu first met Joseph and KIHUMBE during a round table in mid-2017, where basic health data needs and challenges were discussed. Following this, dLab and Data Zetu delivered a basic data training in Mbeya for health and HIV stakeholders, which Joseph attended. This training included some basic data quality practices as well as practical ways to use tools like Excel to record, digitize, and analyze data.

The key to this process was what happened afterwards. In the months following the training, Joseph’s activities at KIHUMBE to foster improved data use and quality involved:

• Raising awareness on the importance of program’s data quality to all staff;
• Using the ordinary feedback sessions platforms and support supervision trips (on monthly basis) to orient the 240 volunteers on data quality skills and tools he learned at the training;
• Putting in place proactive processes to make quality control efforts a part of day-to-day activities across the organization;

Outcomes and Impact

Joseph was one of the 76% of Mbeya training participants who reported an increase in their ability to access, clean, or share data after the training. This has led to several concrete outcomes from KIHUMBE and the people they serve:

• Today, 80% of reports coming from the community volunteers have no any data discrepancies.
• KIHUMBE has been able to more swiftly report to the Water Reed Program, who in turn report back to PEPFAR. This has enabled more responsive resource allocation and adaptive program management.
• Unlike earlier reports, the quarterly report submitted in Dec 2017, had ZERO data discrepancies and without any query from the Water Reed project’s heads. (Usually a quarterly report is delayed for 3-7 days because of the back and forth calls with community volunteers for clarifications, and they’re returned twice or thrice just because of the data quality.)
• As a result of delivering a clean report in the last quarter and capacity in engaging with data, Walter Reed supplied a desktop computer to help KIHUMBE continuing maintaining a good practice in data management.

“The knowledge I have gained helps me to identify data quality issues at earliest stage and be able to rectify before it is being shared with the higher authorities. The daily or weekly reports … now [contain] rigorous formats which make it easier to report on time”.

Joseph, M&E Officer, KIHUMBE

Key Collaborators